Sensors 2026 , 26 , 2049
3of 16
pattern-recognition system, capable of recognising simple and complex odours”. Although there is a tendency to extend the electronic nose term to gas chromatographs and mass spectrometers [11,12], we will use it in the Gardner and Bartlett framework. The array of chemical sensors is the most economical alternative to such instruments, and although it cannot identify the sensed compounds or a human sense of odor [12], it allows the classifica- tion of samples based on the signals obtained from volatiles. Besides detecting the volatiles that interact with their chemical coatings, sensor responses depend on their intensity, which is very difficult to evaluate with a human panel, as sensitivities to different odorants vary, with some individuals exhibiting particularly severe anosmia to specific molecules. Sensing systems, comprising not only the sensors but the sample introduction system, in this case a septum for SPME fiber insertion, a small oven for sample desorption, and the data acquisition and recording system, are designed. Major sensor drawbacks are baseline drift and sensor aging. Baseline drift may be minimized by precisely controlling the temperature and flow rate. Flow injection, which consists of injecting the sample into a carrier flow, allows signal acquisition to take place typically more or less 1 s after the baseline reading. A sensor’s lifetime depends largely on the chosen coatings, and on the temperature and humidity, and may severely limit their use and therefore the number of analyses it can perform. In this work, samples of office papers with 70 to 80 gsm, bought in markets from Germany, Spain, Italy, France, the Netherlands, Russia, Dubai, and Latin America were analyzed. The manufacturer of each office paper sample was known, but the country of production was not always certain, and in some cases the tree species from which the fibers were extracted was also unknown. An efficient SPME extraction was used for sampling, after optimizing the time of exposure and the number of paper discs, as well as the nitrogen flow for desorption and the sample carrier for the new e-nose analytical system. The same sampling method- ology was used for the Gas Chromatography–Mass Spectrometry (SPME-GC-TOF-MS) analysis of volatiles emitted from paper. pH analyses of the samples were performed as a complementary technique. The number of sensors in the array was kept to the minimum amount necessary for paper discrimination. A reduced number of sensors should be an array of the most stable highly sensitive coated crystals that the authors succeeded in preparing. In addition, each selected sensor should respond with a different sensitivity pattern to the emitted VOCs. Colinear sensors, meaning sensors showing similar selectivity and related responses, should not be added to the array. These redundant sensors would not be helpful for paper discrimination, and would most probably contribute to increase noise, and lead to overfitting. A high number of sensors is also a more expensive system, with more integration issues, an increased risk in terms of sensor malfunctioning, and increased difficulties in simultaneous data acquisition and handling. Nevertheless, analysts must always respect the lifetime of SPME fibers and sensors. As raw sensor data are used in this system, without any calibration, which is mandatory and frequent in quantitative analysis, a previously analyzed sample was re-analyzed from time to time to ensure that sensor sensitivities remained the same. The results allowed discrimination between office paper samples. It was possible to dif- ferentiate among paper made from birch, acacia, and Eucalyptus globulus trees. Besides wood origin, a few compounds have been identified as possible markers of geographical origin.
https://doi.org/10.3390/s26072049
Made with FlippingBook interactive PDF creator